Linear fluorocoumarins, also known as psoralens, have been used in combination with ultraviolet light for centuries in cosmetics and for the treatment of proliferative skin diseases such as, for example, vitiligo, eczema, mycosis fungoides, and psoriasis. Terms such as photosensitization, photochemotherapy, photopheresis and PUVA (psoralens ultra violet A radiation) are commonly used to refer to such methods. Recently it was discovered that by modifying the administration of psoralen and ultraviolet light to an offending condition, psoralens can be used to treat cancer (e.g., T cell lymphoma), autoimmune diseases, and microbial infection.
The basic structure of psoralen, with the ring numbering structure used throughout the specification, is shown below: ##STR3##
All psoralens contain two photo-activatable functions (absorbing in the UVA range)--an aryl-conjugated unsaturated pyrone (the coumarin portion) and an aryl-conjugated vinyl ether (the furan portion). All of the commercially available psoralens are highly lipophilic, non-nitrogenous, uncharged small molecules with minimal water solubility. Commercial psoralens are used in over-the-counter cosmetic creams, prescription pharmaceuticals, and as investigational candidates for many of the uses described above. The commercial psoralens in cosmetic/medical use include methoxsalen (also known as xanthotoxin, 8-methoxypsoralen or 8-MOP), trisoralen (also called 4,5',8-trimethylpsoralen, TMP, or trioxsalen), and bergaptan (alternatively named 5-methoxypsoralen or 5-MOP).
The phototherapeutic action of psoralens has been discussed for example, by J. E. Hearst, "Photochemistry of the Psoralens," Chemical Research in Toxicology, 2, 69, 1989 and T. F. Anderson and J. J. Vorhees, Annual Reviews of Pharmacol. and Toxicol., vol. 10, p. 177, 1982. According to these articles, the highly lipophilic psoralens penetrate the target cell's membrane, intercalate into nuclear DNA, and photo crosslink the double helix through bis-cyclobutanes generated from the 3,4-double bond and the 4',5'-double bond [see numbering shown above] to double bonds in DNA's pyrimidine bases. Thus, because the crosslinked DNA is unable to uncoil and function as a template for new gene expression, the target cell is rendered non-viable.
A severe limitation to the acceptance of psoralen-based photochemotherapy or cosmetic skin pigment enhancement, however, is the risk of genetic mutations induced by DNA damage since the natural cellular level repair processes of bi-functional DNA-crosslinks are highly error-prone. Errors in cellular repair processes of true crosslinks translate to mutagenic/carcinogenic events and, in the clinical use of psoralens, represent a significant post-treatment risk of cancer. See, for example, R. S. Stern et al, "Cutaneous Squamous-cell Carcinoma in patients treated with PUVA," New England J. of Med., 1984, pp. 1156-116; R. S. Stern et al, "Malignant Melanoma in Patients Treated for Psoriasis with Methoxsalen and Ultraviolet A Radiation (PUVA)," New England Journal of Medicine, vol. 336, 1997, pp 1041-1045; and W. L. Morrison et al. "Consensus Workshop of the Toxic Effects of Long-Term PUVA Therapy," Arch. Dermatol., vol. 134, 1998, pp. 595-598.
The use of nonlinear furocoumarins (known as angelicins) for the treatment of psoriasis and other skin diseases is taught, for example by U.S. Pat. No. 4,312,883. According to the patent, nonlinear furocoumarins are an effective photochemotherapeutic compounds that does not have the risks associated with psoralens. Nonlinear furocoumarins, however, are limited by their structural geometry, forming only non-crosslinked mono-adducts which have diminished mutagenic behavior. See, for example, R. S. Cole, "Repair of DNA Containing Interstrand Crosslinks in E. Coli," Proc. Nat. Acad. Sci., volume 70, 1973, p. 1064. Further, lipophilic linear psoralens, capable of forming only monoadducts, can be phototoxic to malignant cells. See J. VanDongen, N. D. Heindel et al., "Synthesis of Psoralen Analogs and Evaluation of their Inhibition of Epidermal Growth Factor Binding," J. Pharm. Sci., volume 80, No. 7, July 1991, pp. 686-689.
Despite such risks, an alternative mechanism exists, not involving DNA, by which psoralens can act as phototoxins to a cell. A 22 kDa receptor protein present on psoralen-sensitive cells has been identified as a binding site for photo-activated psoralens. Binding a psoralen to this non-nuclear receptor follows UVA light activation of the psoralen and blocks subsequent binding of epidermal growth factor (EGF) to that receptor. The existence of this non-nuclear target has been described in J. D. Laskin et al., "A Possible Mechanism of Psoralen Phototoxicity Not Involving Direct Interaction with DNA," Proc. Nat. Acad. Sci., vol. 82, pp. 6158-6161, September 1985.
U.S. Pat. Nos. 5,473,083 and 5,216,176 report that reduced and quaternized psoralens are valuable photo-activated therapeutics. Although promising as therapeutics, these dihydro quaternary compounds have often been extremely difficult to synthesize. Furthermore, the reported method of synthesis does not permit access to 5'-(N-pyridiniummethyl) psoralens. 5'-N-pyridiniummethyl)psoralens had been found (in the related fully unsaturated psoralens) to be potent members (active at the lowest concentration levels) of the fully unsaturated psoralens. See Table 1, U.S. Pat. No. 5,216,176. Thus, a need exists for a new synthetic route suitable for 4',5'-dihydro psoralens bearing pyridinium, alkyl amino, alkyl ammonium, or other nitrogen-heterocyclic groups at the 5'-methyl locus.
U.S. Pat. No. 5,473,083 reports the synthesis of 5'-bromomethyl and 5'-quaternary ammonium psoralens (2, with T.dbd.Br or R.sub.3 N+). But, as Scheme 1 below shows, hydrogenolysis of the pendant leaving group-to-carbon bond, not hydrogenation of the 4',5'-double bond, is the dominant outcome when catalytic hydrogenation or exchange hydrogenation are attempted. This unfortunate consequence leads to the recovery of the parent psoralen. ##STR4##
U.S. Pat. No. 5,356,929 discloses the preparation of less labile (to hydrogenolysis) tertiary aminomethylpsoralen. Reduction and subsequent methylation can produce methylated quaternary compounds in low overall yields (35-40%). This is shown in Scheme 2. ##STR5##
Obviously, however, niether 5'-N-pyridiniummethyl 4',5'-dihydropsoralens--nor any quaternary psoralens derived from aromatic heterocyclic amines--are available by this route. Furthermore, as mentioned above, hetero-aromatic quaternary amine groups have been shown to possess beneficial properties in fully unsaturated psoralens. Thus, a need exists for a facile synthetic route to manufacture such quaternary amine-containing 4',5'-dihydropsoralens.
Manufacturing the commercial psoralen, trioxsalen (1) (4,5',8-trimethylpsoralen or TMP), involves alkene bromination in chloroform of the unsaturated allylic bond in 4,8-dimethyl-6-allyl-7-acetoxycoumarin (3, R.dbd.H, R'=Ac). However, if the 7-hydroxyl is not acetylated prior to bromination of the allyl moiety, bromination on C#3 occurs simultaneously with addition of the bromine to the double bond yielding 3-bromo-6-(2,3-dibromopropyl)-4,8-dimethyl-7-hydroxycoumarin (4, R.dbd.Br, R'=H). ##STR6##
In the Kaufman process, the 6-(2,3-dibromopropyl)-4,8-dimethyl-7-acetoxycoumarin (4, R.dbd.H, R'=Ac) is ring-closed by uncapping the 7-acetoxy with sodium ethoxide with concomitant cyclization and double dehydrobromination yielding TMP in 48% yield. See K. D. Kaufman, J. Org. Chem., 1961, 26, 117-121. This reaction involves the loss of two molar equivalents of HBr. Theoretically, this might occur in two separable distinct steps in which one HBr loss is followed by a second through a possible intermediate such as 4,8-dimethyl-4',5'-dihydro-5'-bromomethyl-psoralen (5, R.dbd.H, T.dbd.Br). ##STR7##
As noted above, this theoretically possible intermediate is the very target compound which could not be obtained despite numerous attempts by catalytic hydrogenation or exchange hydrogenation to reduce 4,8-dimethyl-5'-bromomethylpsoralen to 4,8-dimethyl-4',5'-dihydro-5'-bromomethylpsoralen; hydrogenolysis to (1) being the dominant outcome. Alas, careful duplication of Kaufman's synthesis revealed no trace of 4,8-dimethyl-4',5'-dihydro-5'-bromomethyl-psoralen (5, T.dbd.Br), even as a potential transient. NMR analysis of commercially available TMP prepared by Kaufman's route similarly showed none of this 4',5'-dihydro-5'-bromomethyl material as a contaminant. A ready synthesis of this and other valuable intermediates to prepare quaternary ammonium 4',5'-dihydropsoralens remains as a currently unmet need in the art.